In [1]:
import numpy as np 
import pandas as pd
import os
import matplotlib.pyplot as plt
from matplotlib.ticker import FormatStrFormatter, StrMethodFormatter, FuncFormatter
from wordcloud import WordCloud, STOPWORDS
In [2]:
df_top50 = pd.read_csv('D:\PyCharmProject\Top_Youtube_News_Media\_Top50_viewed_video_from_each_channels.csv')
df_stats =pd.read_csv('D:\PyCharmProject\Top_Youtube_News_Media\Mainstream_Media_Statistics.csv')
In [3]:
df_stats.head()
Out[3]:
Unnamed: 0 channelName id subscribers total views total videos created date description playlistId
0 0 Fox News UCXIJgqnII2ZOINSWNOGFThA 10500000 14839464315 93767 2006-09-19T01:48:52Z FOX News Channel (FNC) is a 24-hour all-encomp... UUXIJgqnII2ZOINSWNOGFThA
1 1 Al Jazeera English UCNye-wNBqNL5ZzHSJj3l8Bg 10500000 3343817389 105487 2006-11-23T14:12:43Z #AlJazeeraEnglish, we focus on people and even... UUNye-wNBqNL5ZzHSJj3l8Bg
2 2 Vox UCLXo7UDZvByw2ixzpQCufnA 11400000 3260217205 1593 2014-03-04T20:30:22Z Vox helps you understand our world.\n\nVox vid... UULXo7UDZvByw2ixzpQCufnA
3 3 Good Morning Britain UCq18eeL7D9Vd8DhjMcLh9QQ 1200000 1048952422 11212 2014-08-18T09:23:49Z The Good Morning Britain YouTube channel deliv... UUq18eeL7D9Vd8DhjMcLh9QQ
4 4 E! News UCjDsbbzHgTrGc4Ff26TJtsA 1220000 515416062 12501 2005-10-25T11:21:48Z E! News is your go-to source for breaking cele... UUjDsbbzHgTrGc4Ff26TJtsA
In [4]:
df_stats.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 43 entries, 0 to 42
Data columns (total 9 columns):
 #   Column        Non-Null Count  Dtype 
---  ------        --------------  ----- 
 0   Unnamed: 0    43 non-null     int64 
 1   channelName   43 non-null     object
 2   id            43 non-null     object
 3   subscribers   43 non-null     int64 
 4   total views   43 non-null     int64 
 5   total videos  43 non-null     int64 
 6   created date  43 non-null     object
 7   description   41 non-null     object
 8   playlistId    43 non-null     object
dtypes: int64(4), object(5)
memory usage: 3.1+ KB
In [5]:
df_stats['channelName'].unique()
Out[5]:
array(['Fox News', 'Al Jazeera English', 'Vox', 'Good Morning Britain',
       'E! News', 'ABC News In-depth', 'BBC News', 'ABC News',
       'CBC News: The National', 'CNN', 'ITV News', 'On Demand News',
       'AFP News Agency', 'CBS Evening News', 'SABC News',
       'Forbes Breaking News', 'The Guardian', 'Washington Post',
       'euronews', 'CNBC Television', 'NowThis News', 'Daily Mail',
       'Voice of America', 'Channel 4 News', 'The Economist', 'CBC News',
       'Sky News Australia', 'Global News', 'PBS NewsHour', 'USA TODAY',
       'The Telegraph', 'The Sun', 'The New York Times', 'PINAS INSIDER',
       'Wall Street Journal', 'DW News', 'CBS News', 'The Young Turks',
       'MSNBC', 'Complex', 'Sky News', 'WION', 'NBC News'], dtype=object)
In [6]:
df_stats['channelName'].duplicated().any()
Out[6]:
False
In [7]:
df_stats.isna().sum()
Out[7]:
Unnamed: 0      0
channelName     0
id              0
subscribers     0
total views     0
total videos    0
created date    0
description     2
playlistId      0
dtype: int64
In [9]:
df_stats=df_stats.fillna('N/A')
In [10]:
#colors for bar graphs and functions for y axis labels
cmap = plt.cm.tab10
colors = cmap(np.arange(10) % cmap.N)

def millions(x, pos):
    return '%1.0fM' % (x * 1e-6)

formatterm = FuncFormatter(millions)

def billions(x, pos):
    return '%1.0fB' % (x * 1e-9)

formatterb = FuncFormatter(billions)
In [11]:
#Get df based on first feature arg
def get_relation_mul(df,top_n,*col):
    df_feature=df.filter([*col])
    feature=[*col][0]
    df_feature= df_feature.sort_values(feature, ascending=False)
    df_feature=df_feature.nlargest(top_n,feature)
    return df_feature
In [12]:
df_stats_views=get_relation_mul(df_stats,10,'total views','channelName','subscribers','total videos')
df_stats_views['views/sub']=df_stats_views['total views']/df_stats_views['subscribers']
df_stats_views['views/video']=df_stats_views['total views']/df_stats_views['total videos']

fig,(ax1,ax2,ax3) = plt.subplots(3,1, figsize=(10,20))

fig.tight_layout(h_pad=15.0)

ax1.bar(df_stats_views['channelName'],df_stats_views['total views'], color=colors)
ax2.bar(df_stats_views['channelName'],df_stats_views['views/sub'], color=colors)
ax3.bar(df_stats_views['channelName'],df_stats_views['views/video'], color=colors)

ax1.set_xlabel('Channel')
ax1.set_ylabel('Total Views')
ax1.set_title('Top 10 Viewed Channels')
ax1.yaxis.set_major_formatter(formatterb)

ax2.set_xlabel('Channel')
ax2.set_ylabel('Views per Sub')
ax2.set_title('Views per subscriber')

ax3.set_xlabel('Channel')
ax3.set_ylabel('Views')
ax3.set_title('Views per video')
ax1.tick_params(axis='x', labelrotation = 90)
ax2.tick_params(axis='x', labelrotation = 90)
ax3.tick_params(axis='x', labelrotation = 90)
In [13]:
df_stats_videos=get_relation_mul(df_stats,10,'total videos','channelName','subscribers','total views')
In [14]:
df_stats_videos['videos/sub']=df_stats_videos['total videos']/df_stats_videos['subscribers']
df_stats_videos['views/video']=df_stats_videos['total views']/df_stats_videos['total videos']
In [15]:
fig,(ax1,ax2,ax3) = plt.subplots(3,1, figsize=(10,20))
fig.tight_layout(h_pad=15.0)

ax1.bar(df_stats_videos['channelName'],df_stats_videos['total videos'], color=colors)
ax2.bar(df_stats_videos['channelName'],df_stats_videos['videos/sub'], color=colors)
ax3.bar(df_stats_videos['channelName'],df_stats_videos['views/video'], color=colors)
ax1.set_xlabel('Channel')
ax1.set_ylabel('Total Videos')
ax1.set_title('Channels with Most Videos')
ax2.set_xlabel('Channel')
ax2.set_ylabel('Videos per Sub')
ax2.set_title('Videos per subscriber')
ax3.set_xlabel('Channel')
ax3.set_ylabel('Views/video')
ax3.set_title('Views per Video')
ax1.tick_params(axis='x', labelrotation = 90)
ax2.tick_params(axis='x', labelrotation = 90)
ax3.tick_params(axis='x', labelrotation = 90)
In [16]:
df_stats_subs=get_relation_mul(df_stats,10,'subscribers','total views','total videos','channelName')
In [17]:
df_stats_subs['views/sub']=df_stats_subs['total views']/df_stats_subs['subscribers']
df_stats_subs['sub/video']=df_stats_subs['subscribers']/df_stats_subs['total videos']
In [18]:
fig,(ax1,ax2,ax3) = plt.subplots(3,1, figsize=(10,20))
#fig.suptitle('Top 10 Channels with most videos')
fig.tight_layout(h_pad=15.0)

ax1.bar(df_stats_subs['channelName'],df_stats_subs['subscribers'], color=colors)
ax2.bar(df_stats_subs['channelName'],df_stats_subs['views/sub'], color=colors)
ax3.bar(df_stats_subs['channelName'],df_stats_subs['sub/video'], color=colors)
ax1.set_xlabel('Channel')
ax1.set_ylabel('Total Subscribers')
ax1.set_title('Top 10 Subscribed Channels')
ax1.yaxis.set_major_formatter(formatterm)
ax2.set_xlabel('Channel')
ax2.set_ylabel('Views per Sub')
ax2.set_title('Views per subscriber')
ax3.set_xlabel('Channel')
ax3.set_ylabel('Subscribers per video')
ax3.set_title('Subscribers per video')
ax1.tick_params(axis='x', labelrotation = 90)
ax2.tick_params(axis='x', labelrotation = 90)
ax3.tick_params(axis='x', labelrotation = 90)
In [19]:
stopwords=set(STOPWORDS)
In [20]:
def generate_wordcloud(words, title, max_words=25, figsize=10):
    wordCloud=WordCloud(background_color='white',
                    scale=5, 
                    min_font_size=5,
                    max_words=max_words,
                    stopwords=stopwords).generate(words)
    plt.figure(figsize=(figsize,figsize))
    plt.axis("off") 
    plt.title(title, pad=20, fontsize=20)
    
    plt.imshow(wordCloud) 
    plt.show()
In [21]:
generate_wordcloud(str(df_stats['description']),"WordCloud of Channel Descriptions")
In [22]:
df_top50.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2200 entries, 0 to 2199
Data columns (total 11 columns):
 #   Column        Non-Null Count  Dtype  
---  ------        --------------  -----  
 0   Unnamed: 0    2200 non-null   int64  
 1   Video Id      2200 non-null   object 
 2   channelTitle  2200 non-null   object 
 3   title         2200 non-null   object 
 4   publishedAt   2200 non-null   object 
 5   categoryId    2200 non-null   int64  
 6   description   2198 non-null   object 
 7   viewCount     2196 non-null   float64
 8   likeCount     2186 non-null   float64
 9   commentCount  1964 non-null   float64
 10  duration      2200 non-null   object 
dtypes: float64(3), int64(2), object(6)
memory usage: 189.2+ KB
In [23]:
df_top50.head()
Out[23]:
Unnamed: 0 Video Id channelTitle title publishedAt categoryId description viewCount likeCount commentCount duration
0 1400 spPo80dPUwI USA TODAY Heroic mom saves terrified five-year-old from ... 2022-12-05T17:28:31Z 25 A mother came to her five-year-old daughter's ... 347180823.0 10370485.0 269937.0 PT41S
1 1900 NxRGRFKBVLg On Demand News Fans Catch CAT Plummeting from Stadium 2021-09-13T15:02:56Z 25 'Fans Catch CAT Plummeting from Stadium'\n\nFa... 121244026.0 6269691.0 95053.0 PT44S
2 1401 KDo8n_7_2YI USA TODAY Brave woman fights off male attacker while alo... 2023-02-17T13:28:54Z 25 A 24-year-old woman successfully fought off a ... 112886052.0 4248640.0 92533.0 PT37S
3 1402 _Gtm2j2gb6w USA TODAY Hurricane Fiona devastates Puerto Rico, intens... 2022-09-22T17:29:38Z 25 Hurricane Fiona flooded Puerto Rico with 6 to ... 97966113.0 1504843.0 7100.0 PT42S
4 650 or_MDJnmihc CBS News Defendant collapses in court after guilty verdict 2017-11-14T17:36:13Z 25 Diana Lovejoy collapsed in a California courtr... 75970739.0 371733.0 79515.0 PT2M31S
In [24]:
generate_wordcloud(str(df_top50['description']),"WordCloud of Top 50 videos description from each channel", max_words=50)
In [25]:
generate_wordcloud(str(df_top50['title']),"WordCloud of video Titles", max_words=50)
In [27]:
list_sub=df_stats_subs['channelName'].values.tolist()
In [28]:
#WordCloud for top 10 subscribed channels videos
for ch in list_sub:
    df_top50_ch=df_top50.loc[df_top50['channelTitle']==ch]
    generate_wordcloud(str(df_top50_ch['description']),f"WordCloud of top subbed channels: {ch}", max_words=50)
In [29]:
list_videos=df_stats_videos['channelName'].values.tolist()
In [30]:
#Wordcloud of top 10 channels with most videos
for ch in list_videos:
    df_top50_ch=df_top50.loc[df_top50['channelTitle']==ch]
    generate_wordcloud(str(df_top50_ch['description']),f"WordCloud of channels with most videos: {ch}", max_words=50)
In [213]:
list_views=df_stats_views['channelName'].values.tolist()
In [214]:
for ch in list_views:
    df_top50_ch=df_top50.loc[df_top50['channelTitle']==ch]
    generate_wordcloud(str(df_top50_ch['description']),f"WordCloud of channels with most views: {ch}", max_words=50)
In [31]:
df_top50_ch_list=df_top50['channelTitle'].unique().tolist()
In [32]:
#Get top N videos of each channel based on feature
def get_top_channel_feature(feature1, feature2, top_n):
    description_list=[]
    for ch in df_top50_ch_list:
        df_top50_ch=df_top50.loc[df_top50['channelTitle']==ch]
        df_top50_ch=df_top50_ch.sort_values(feature1, ascending=False)
        for n in range(top_n):
            description=df_top50_ch.iloc[n-1][feature2]
            description_list.append(description)
    return description_list
In [33]:
description_list_view=get_top_channel_feature('viewCount','description',2)
In [34]:
generate_wordcloud(str(description_list_view),"WordCloud of most viewed videos for each channel", max_words=50)
In [35]:
description_list_like=get_top_channel_feature('likeCount','description',2)
In [36]:
generate_wordcloud(str(description_list_like),"WordCloud of most liked videos for each channel", max_words=50)
In [ ]: